Model instabilities etc …

[stephane]

Hello,

I would like to share with you my latest thoughts & questions regarding the optimization of hydraulic models with HECRAS.

Let’s take an example :
A classic full 2D network of 150,000 cells (size between 2 m and 10 m) for which I defined a simulation of duration 4 hours at a time step of 1 second. The maximum iterations is set to 20 and the water surface tolerance is 0.01 m. The calculations are fully dynamic.

If everything goes well and there are no errors, HECRAS therefore performs 4h x 60min x 60sec times the calculation of the hydraulic characteristics of the 150,000 cells (therefore in total 14,400 times).

If I have instabilities at each time step (the worst case scenario), HECRAS therefore performs a maximum of 20 x 4h x 60min x 60sec times the calculation of the hydraulic characteristics of the 150,000 cells (therefore 288,000 times).

In fact, after the calculations ends, I got 20870 compute messages on more than thirty different cells spread over the network. The instabilities varied between 0.015 m and 0.125 m.

First question :
Each time HECRAS detects an instability and displays a compute message, did it repeat the calculation 20 times (as in my example) on the whole network or only on the cell which is unstable, before display a calculation error? A bit like a matrix calculation …

Second question:
Why can’t we still view the position of these cells directly on the network map in RASMapper today? Is it complicated to go from this sequential list of error messages (difficult to use) to a much interesting map to optimize the model?

Third question:
Is it important to focus on these messages if the instabilities remain within acceptable values ​​which do not call into question the overall validity of the model?

[cameron]

First question – I believe it does it for the entire network

Second question – I think this to myself every single time I start to optimize the model and fix the cells manually. I would email HEC-RAS and tell them to add this to Mapper.

Third – I look at a few things when it comes to these instabilities:
1. Where are they occurring. If they are outside of my area of interest I focus less on fixing them
2. What is the volume error. If the volume error is above a certain threshold, then I try to fix them. This threshold varies from project to project.
3. When are they occurring. If they happen before or after the peak and won’t have an impact on the max floodplain (if that is what I care about), then I may not fix them.
4. How often they occur for certain cells. I will check to see how many iteration issues occur for each grid cell and if it is above a certain threshold, i will fix them. This can help speed up the runs significantly.

It is also important to check the total volume error in deciding if further adjustments are needed.

[Author]

Posts